The just-mcp server defines 4 tools with complete input schemas and basic descriptions. All tools have proper naming with action verbs (list_, run_, get_, validate_). Descriptions are present but generic, most fall below the 194-char production baseline. Parameter schemas are well-typed (string, boolean) with descriptions, but lack detail about ranges, formats, and valid values. Output schemas are not documented in the visible code. Error handling is present (errorReporting=true) but recovery guidance is not visible. The tool set is logically composed with minimal overlap. Naming is clear: list_recipes, run_recipe, get_recipe_info, validate_justfile, each is unambiguous and verb-led.

• Expand each tool description to 100-200 chars, including WHAT it does, WHEN to use it (vs a similar tool), and what structure it returns. E.g., 'List all recipes defined in the justfile (or specified path). Returns array of recipe names and their descriptions. Call this first to explore available commands before executing one.'
• Document output schemas for all tools. E.g., 'Returns {recipes: [{name: string, description: string, args: array<string>}], total: number, hasMore: boolean}'
• Add limit and offset parameters to list_recipes with docs: 'Maximum recipes to return (1-100, default 20). Use with offset for pagination.' Return total_count and next_offset in response.
• For run_recipe.args, specify: 'JSON array of string arguments, e.g. ["--quiet", "--verbose"]. Passed as-is to the recipe command.'
• Add state-modification warning to run_recipe: 'Executes a recipe, which may modify files, environment, or external services. This action is irreversible.'
• Consider adding an optional dry_run parameter to run_recipe: 'If true, parse recipe and show what would execute without actually running it. Helps agents verify intent before destructive operations.'
• Document validate_justfile output: what errors does it return? Structure: {valid: boolean, errors: [{line: number, message: string}]?}
• Add parameter to all tools: 'working_dir (optional string): Directory from which to run the recipe. Defaults to current working directory.' This lets agents execute recipes in specific contexts.